Two Pensions, Two Social Security Records, and a Widow's Penalty

Start with the survivor benefit, because that is what the claiming decision is really about. When one spouse dies, the survivor keeps the larger of the two benefits, not both. The survivor benefit is based on what the deceased was receiving, including delayed retirement credits. If Alan claims at 67 and dies first, Beth's survivor benefit is $3,120; if he waits to 70, it is $3,120 times 1.24, or $3,869. With a six-year age gap, excellent health on Beth's side, and longevity into the nineties in her family, the probability that Beth is a widow for a long stretch is high. Delaying Alan's claim is effectively buying the best available longevity insurance for the household.

Then handle Beth's own record. Her full retirement age benefit of $1,460 is just under half of Alan's PIA of $3,120, which is $1,560. That means she is eligible for a spousal top-up of about $100 a month at her full retirement age once Alan has filed. That is small, so it should not drive the decision.

Practical sequence:

• Beth claims her own reduced benefit at 62, around $1,020, to provide cash flow while Alan delays. Her own benefit is permanently reduced, but the reduction does not carry into her survivor benefit, which steps up to Alan's amount.
• Alan files at 70, not at 67 and not at 64 when he stops working. Between 64 and 70 the household bridges the gap from the taxable account and cash, spending roughly $37,000 a year of Alan's forgone benefit from a portfolio that can afford it.
• Note that deemed filing does not force a spousal application here because Beth is filing on her own record and both were born after 1954.

Model the break-even honestly. Alan's delay from 67 to 70 forgoes about $112,000 of benefits to add roughly $749 a month, an index-adjusted payback of about 12.5 years for him alone, and considerably faster when weighted by the chance Beth survives him. Framing it as a joint-life decision rather than an individual break-even is what makes the recommendation defensible.

Compute the implied payout rates. Single life: $1,850 times 12 is $22,200, or 6.57% of $338,000. Joint and 100% survivor: $1,590 times 12 is $19,080, or 5.64%. In today's rate environment those are competitive with what a conservative portfolio could safely support, and they carry no investment risk, no sequence-of-returns risk, and no longevity risk.

Weigh the individual factors:

• Health and longevity. Alan's father died at 72, which cuts against a single-life annuity; Beth's family history runs to the nineties, which strongly favors the joint option. The joint and survivor selection protects exactly the risk that matters, which is Alan dying early and Beth living thirty more years on a reduced income.
• Insurer and plan risk. Check whether the plan is private or covered by the Pension Benefit Guaranty Corporation and at what guaranteed maximum; a lump sum removes plan risk but transfers it to market and longevity risk.
• The lump sum is rollable to an IRA, which preserves flexibility for Roth conversions and for heirs. That flexibility has real value, but it must be weighed against the loss of a guaranteed inflation-relevant income floor.
• Household funding. Total spending of $92,000 after tax against Social Security plus a joint pension of $19,080 plus Alan's eventual $46,400 a year covers most of the target with guaranteed income, which sharply reduces the withdrawal rate required from the $1.15M portfolio.

Recommendation. Take the joint and 100% survivor annuity if the plan is soundly funded, funded status should be verified in the summary annual report, and if no survivor option is needed later then revisit. The marginal cost of the survivor protection, $260 a month, is cheap insurance against a 30-year widowhood. If liquidity or an early death before the payback period is a dominant concern, split the decision by taking the joint annuity and preserving the Roth accounts for flexibility.

Document the assumption set, especially the discount rate implied by the plan's lump-sum calculation, since plans typically use a segment-rate approach that moves with interest rates.

Identify the window. Alan was born in 1962, so under current law his required beginning date is age 75, not 73. That gives roughly eleven years from age 64 with no wage income and before required distributions begin, which is the cheapest time the household will ever have to move money out of the traditional IRAs.

Quantify the problem being solved. Traditional IRA balances are $480,000 today and will grow. More important than the balance is the filing status shift: when Alan dies, Beth files single, with roughly half the bracket widths and roughly half the income thresholds for Medicare surcharges, on an income that will have barely changed. That widow's penalty is the strongest argument for converting now while the married-filing-jointly brackets are available.

Build the annual conversion plan:

• Convert enough each year to fill the 24% bracket, but not so much that Medicare surcharges are triggered unnecessarily.
• Remember the two-year lookback. Medicare income-related premium surcharges for a given year are based on modified adjusted gross income from two years earlier, and Roth conversions count in that figure. A large conversion at age 64 affects premiums at 66; a conversion at 65 or later affects premiums directly.
• Pay the conversion tax from the taxable account or cash, never from the converted amount, so the full balance lands in the Roth.
• Check state tax, and consider whether a future move to a lower-tax state argues for waiting.
• Leave enough in traditional IRAs to use the standard deduction and the lower brackets in later years, and to fund charitable qualified charitable distributions once eligible at 70 and a half.

Sequence around other income. Do the largest conversions in the years before Social Security begins at 70 and before the pension's tax character is fully absorbed; after 70 the guaranteed income floor consumes bracket space quickly.

Set a guardrail. Recalculate each November using actual year-to-date income, and execute the conversion in December rather than January, so the amount is a measured decision rather than a guess.